AN IMPROVED KERNEL ISOMAP ALGORITHM WITH APPLICATION TO IMAGE RETRIEVAL

The conventional kernel ISOMAP algorithm (K-ISOMAP) can not work well in keeping the intrinsic topology of datasets from multi-class clusters datasets in the low-dimensional space. In order to avoid this shortcoming, a novel algorithm named kernel multi-class multi-manifold ISOMAP (K-MCMM-ISOMAP) is proposed in this paper, which is the kernel version of MCMM-ISOMAP. The new algorithm doesn't only discover intrinsic topology of data in low-dimensional mapping space, but also can explicitly generated the mapping function from high-dimensional space to low-dimensional space. Therefore it can be applied into the image retrieval system consisting of multi-class image dataset. The experimental results show that the new algorithm is more effective than the K-ISOMAP.